Three Repeated Intrathecal Injections of Allogenic Umbilical Cord-Derived Mesenchymal Stromal Cells in Autism Spectrum Disorder: A Phase I Clinical Trial

Major Points and Findings:

Of the studies published in the last year, this is the one that most closely matches the way cells are delivered in our own autism protocol. Umbilical cord-derived cells were given intrathecally, more than once. With five patients it is a safety study and cannot show whether the treatment works. The paper does describe the procedure and the eligibility criteria in more detail than most.

Aim:

To establish whether three consecutive intrathecal injections of allogeneic umbilical cord-derived mesenchymal stromal cells (UC-MSCs), given a month apart, are safe and feasible in young children with autism and, secondarily, to look for any signal in behaviour, brain imaging, EEG and cerebrospinal fluid inflammation.

Methods:

This was a phase I, open-label, single-arm trial run at the Children’s Medical Center and Roozbeh psychiatric hospital, Tehran University of Medical Sciences, Iran. It was registered as IRCT20200217046526N3.

Ninety-six children were screened between July 2022 and January 2024 and five were enrolled. They were 58 to 72 months old, four of the five male, all with DSM-5 Level II autism and a stable medication regimen. Only five of the ninety-six screened were eligible, which reflects a long and specific exclusion list:

• Age outside 36 to 84 months
• A known genetic syndrome associated with autism (Fragile X, Rett, tuberous sclerosis, neurofibromatosis, PTEN mutation)
• A cerebral palsy diagnosis
• Active seizure disorder within the past six months
• Any prior history of CNS infection or inflammatory disease
• Coexisting psychiatric diagnoses, immunodeficiency, active infection, malignancy, or a history of allergic disease
• Any previous allogeneic tissue transplant

Each child received a fixed dose of 20 × 10⁶ GMP-grade UC-MSCs in 2 mL of saline with 5% human serum albumin. Children were kept nil by mouth for at least eight hours and sedated intravenously with thiopental sodium (3 to 5 mg/kg). A lumbar puncture was performed with a 24-gauge spinal needle at L3/L4 or L4/L5. Three millilitres of CSF were withdrawn. One was kept for analysis and two were re-infused after the cells, so that no cells were left in the dead space of the needle. Children were watched for four hours and discharged if stable, with a clinician on call for the next 48 hours.

The primary outcome was adverse events graded by CTCAE v5. CARS, ATEC and the Behavior Problem Inventory were recorded at baseline and at one, two, four and six months, CSF cytokines before each injection, and EEG and diffusion tensor imaging at baseline and six months. Parents’ own health was also tracked, using GHQ-28 and SF-36. With five participants, the authors used no inferential statistics and report everything as individual observed values.

Results:

Safety: No serious adverse events occurred. Every one of the five children had at least one adverse event, and the events followed a pattern:

• Fever occurred in 5 of 5, vomiting in 4 of 5 and nausea in 3 of 5.
• Fever was Grade 1 in three children and Grade 2 in two. None had meningeal signs such as neck stiffness, photophobia, altered consciousness or a positive Brudzinski’s sign. Every episode settled within 12 to 48 hours on paracetamol at 10 mg/kg.
• Adverse events grew more numerous and more severe with each successive injection. The first injection was the one best tolerated.
• Nothing was reported at the four- and six-month visits, CSF cultures grew nothing, and no laboratory abnormality of consequence appeared.

The abstract lists nausea as the second most common event at 4 of 5. The full text reports vomiting at 4 of 5 and nausea at 3 of 5. This summary follows the full text.

Behaviour: Total CARS scores fell in all five children, from a median of 43 (range 30 to 45) to 31.5 (range 22 to 35), a median change of −26.7%. Two children moved into the non-autistic band and three moved from severe to mild-to-moderate. Most of the change appeared within the first month, after which scores broadly plateaued.

Within the CARS, the items that changed most were activity level, emotional response, visual response, and object and body use. Those that changed least were non-verbal communication, adaptation to change, taste-smell-touch response, imitation and general impression.

The ATEC results were less favourable. The median change was −33.3%, but the two children who were non-verbal at baseline showed minimal to no change in the speech and language subdomain. After the first two months ATEC scores either held steady (three children) or rose again (two children). On the Behavior Problem Inventory the results were mixed. Self-injury improved in three and resolved completely in one, but worsened in another. Aggression increased in two. Stereotyped behaviour increased in three of the five and decreased in two.

Inflammation: CSF pro-inflammatory cytokines fell consistently over two months. IL-6 dropped from a median of 7.66 to 2.49 pg/mL and TNF-α from 3.26 to 1.1 pg/mL. CSF white cell count and CRP also fell. The authors note that there is no validated normal range for CSF cytokines in children, because lumbar puncture cannot ethically be performed on healthy children. The values are therefore within-child changes and cannot be compared against a norm.

EEG and imaging: EEG was attempted in four children and only two recordings were usable. The rest were lost to artefact and non-compliance. In those two, absolute delta power fell 53.5% and theta 20.6%, with alpha up 20.2% and beta up 65.0%. The recordings were made under sedation, which itself pushes EEG toward slower frequencies, so the authors treat before-and-after comparison as valid but any comparison with normative data as invalid. Diffusion tensor imaging showed no significant structural remodelling of the corticospinal tracts or cingulum bundles.

Conclusions:

Three repeated intrathecal injections of UC-MSCs were safe and feasible in five children. The authors state that everything beyond safety is exploratory and hypothesis-generating rather than confirmatory. The limitations they list are the absence of a control group, so that treatment effects cannot be separated from ordinary development, a sample too small to detect uncommon adverse events, an open-label design in which both caregivers and assessors knew what had been given, and no long-term follow-up.

On expectations, they write that it is crucial to maintain a balanced and evidence-based perspective, “ensuring that caregivers and stakeholders have realistic expectations and are protected from misleading claims of a complete cure.”

Background Information:

Two points from the discussion are relevant to this treatment.

The first concerns fever. MSCs have low immunogenicity but, as the authors put it, they lack full immunological privilege. A separate meta-analysis of MSC-related adverse events found fever to be the most common event, particularly with allogeneic cells, and found that when injections were given three times a ten-fold increase in dose significantly raised the likelihood of fever. One earlier study found that a third of patients developed new anti-HLA antibodies after intravenous MSCs. The immunological consequences of repeated dosing are one of the open questions.

The second concerns who is likely to benefit. The authors argue that because the mechanism is immunomodulatory, UC-MSCs are most likely to help the subgroup of children in whom neuroinflammation is a primary driver, and correspondingly less likely to help children whose autism arises mainly from genetic or structural causes. They also point to work by Dawson and colleagues showing that improvement after cord blood infusion tracked with baseline non-verbal IQ, with meaningful gains in children without intellectual disability and not in those with it. Younger children responded better. For future trials they recommend stratification on immunological, demographic and cognitive grounds at baseline, so that likely responders can be identified in advance.

They also observe that where improvement faded in some children, adjunctive behavioural or occupational therapy after cell treatment might prolong it. They present this as a hypothesis to be tested.